Mira4 Assembler using all memory for denovo, est assembly

[nareshvasani]

Hi Users,

My PC has 48 GB of RAM and 8 CPU. I am trying to create denovo assembly using Mira 4 Assembler. Input file has only 2 fastq files with only 30 million reads with sequence length 50-350 nt.

I used below manifest.comf file

project = miraproject job = est, denovo, accurate
parameters = COMMON_SETTINGS -GENERAL:number_of_threads=4 amm=off -HS:mnr=yes:nrr=8
parameters = IONTOR_SETTINGS -CL:qc=no -ALIGN: min_relative_score=90 -ASSEMBLY:minimum_read_length=200

readgroup = SomeUnpairedIonTorrentReadsIGotFromTheLab
technology = iontor
data = /media/DATAPART3/Combine_Files/miraproject/data/


But problem is it is taking long time more than 20 hours and it is using all the memory.

Any suggestion?


I would really appreciate your help.

Thanks in advance.
Naresh

[nareshvasani]

Hi Users,

bump....

[sklages]

You should post the issue on the mira mailing list.

[akorobeynikov]

You may also consider SPAdes 3.0, which contains experimental IonTorrent error correction and assembly module.

[nareshvasani]

SPAdes

Hi akorobeynikov,

Thanks for your reply. I am doing transcriptome analysis and SPAdes is useful for only genome analysis.
Am i right?

Best,
Naresh

You may also consider SPAdes 3.0, which contains experimental IonTorrent error correction and assembly module.

[nareshvasani]

Hi sklages,

I contacted them, they just told me get new computer resources with more GB on it.

Thanks,
Naresh

You should post the issue on the mira mailing list.

[akorobeynikov]

Thanks for your reply. I am doing transcriptome analysis and SPAdes is useful for only genome analysis.
Am i right?

Well, I never did this by myself, but rumors are that you still
may try to assemble them in single-cell mode. I'd select the standard set of kmers, that is - 21,33,55

[danwiththeplan]

Try a less memory-heavy assembler that is based on de bruijn graphs (eg Trinity, transABYSS)? I think Mira is based at least partially on overlap-layout-consensus so will be very memory hungry. If you have 30 million reads, this line from the Mira4 manual may be relevant:

1.4. For which data sets to use MIRA and for which not

As a general rule of thumb: if you have an organism with more than 100 to 150 megabases or more than 20 to 40 million reads, you might want to try other assemblers first.